Imagine being given the task of simultaneously migrating your legacy servers to a common platform, deploying a new generation of thin clients across your network of branches, and integrating an additional 20 per cent of customer data resulting from an acquisition. Wayne Miller, MIS manager for HCF, was faced with this predicament in April 2002 when the company embarked on an enterprise-wide IT infrastructure optimisation project.
The NSW health insurer’s infrastructure spreads outwards from an IBM Z800 mainframe running OS/390 at the centre to some 250 WYSE thin clients at the branches. In the middle are 50 Intel servers running Windows NT. Augmenting the mainframe is a 1.4TB StorageTek SAN and an IBM tape auto-loader, which is used for backups.
“Our objectives for the optimisation project was to replace legacy systems running NetWare to Citrix on Windows NT,” Miller said.
“This project completed the consolidation of all our server operating systems.”
However, according to Miller, that initial objective quickly expanded into an audit of the entire HCF network. “At the time HCF was one of the few companies still running dumb terminals so it was time for us to upgrade,” he said. “Also, as part of the Citrix rollout, we moved our network infrastructure to a Telstra private IP network that runs over frame relay. We considered VPNs but then decided they are inappropriate for our needs. Cost is only one factor when you invest in infrastructure.”
HCF’s infrastructure upgrade coincided with the company’s acquisition of IOR — a smaller health insurer with offices in Queensland, Victoria and South Australia.
“IOR’s transaction system was running HAMBS — a popular health fund transaction system — on Unix before we migrated the data to our mainframe,” Miller said. “The mainframe was also upgraded to handle conversion and data integration.”
HCF processes about 369,000 transactions a day, which was combined with an estimated 60,000 a day from IOR.
“Converting the data was a hard slog,” Miller said. “The two systems have different naming conventions and varying data definitions so you can’t work solely on the name of a particular data field.”
If Miller were to give advice on what to look out for when faced with such a large integration project it would be “get it right with data integrity”.
“The toughest part of the integration project was achieving data integrity. We have several staff who know our data well and worked on it for three months,” he said.
Miller said that each part of the infrastructure optimisation project was subjected to the same disciplined process undertaken with all the company’s spending.
With the large-scale optimisation projects complete, Miller is happy with the outcome.
“Now our staff can access Internet, intranet and e-mail through the thin clients,” Miller said. “The NetWare legacy systems were high maintenance and support for both operating systems was high. Running Citrix thin clients also enables the development of the network to be realised.”
Miller claims that the whole process was transparent to the users and came in under budget, because the preparatory work was done in-house. “Other tangible benefits of the optimisation include, better training and operating procedures, reduced information access overheads such as transaction time, as well as reduced paper and delivery costs,” Miller said. “Our post-implementation review conservatively estimates a 10 per cent increase in overall productivity.”
In addition to the tangible benefits of such an optimisation, Miller also noted how the infrastructure can be applied to future projects.
“There are also many intangible benefits such as a resulting platform for CRM implementations, magnetic swipe card reader support for identity management, and the ability to run Lotus Notes which is less vulnerable than Outlook,” Miller said.
“Overall, we can say that our customer service has been improved and in our industry that’s imperative. For example, membership eligibility through hospitals is now Web-based and runs around the clock. Also, all our member services are dynamically updated in real time.”
Among large organisations, optimisation projects can be applied to specific infrastructure segments in order to help improve overall business performance.
Last year St George Bank established a SAN with the intention of saving storage and server infrastructure running costs.
Phil Soutter, general manager of IT production at the bank, said there’s a number of objectives for the SAN. “Most of our objectives for the SAN have been met, including more effective and efficient use of disk resources, improvement in disk management, improved local and offsite disaster recovery (DR), and a complete re-architecture of the open systems backup and recovery strategies,” Soutter said. “Other improvements are human and hardware savings, and greater systems availability through greatly reduced hardware maintenance windows.”
Approval for the SAN was granted after the regular business requirements, goals and objectives were determined. Once a plan for the project was mapped, the bank spoke with 19 vendors and then short-listed eight. From the eight, four companies were then asked to submit proposals, including costs, and finally two were selected for consideration. Approval by the St George group executive management team was granted after the final solution was designed and a six-week proof of concept was conducted.
“It was well worth doing,” Soutter said. “The SAN is much more efficient, has better disaster recovery as well as business continuity. We have also avoided a much larger investment in additional storage.”
Soutter said all 12 legacy systems scheduled to participate in the initial implementation phase converted onto the SAN by September 2002. This included the production server at Kogarah (south of Sydney) and all DR and development servers in the DR site.
In all 12 systems were transitioned in six months using five different operating systems at two sites without one interruption to the business.
“Re-architecting the open systems — about 1100 servers — to the new backup and recovery software and hardware across Australia and the removal of all the old hardware and software took about six months,” Soutter said. In terms of an ROI for the SAN, Soutter said it was less than 18 months in infrastructure and support cost savings alone.
“One storage administrator is now capable of managing far greater amounts of disk then ever before,” Soutter said. “The applications are running faster and we are also now seeing an extension in the life of our production servers.” Other, less tangible values to St George concern what has been delivered to customers such as improved response time, greater system availability and guaranteed data integrity with no data loss in times of a disaster.